Colorado Decision Support System for Prediction of ...€¦ · Colorado Decision Support System for...

Colorado Decision Support System for

Prediction of Wildland Fire Weather, Fire

Behavior, and Aircraft Hazards

0

Kick-off Briefing

3 December 2015

Fire Prediction Technology Description

The CO-FPS will employ a sophisticated, coupled atmosphere - wildland fire model (e.g., CAWFE®), that generates a high-resolution 4D gridded weather simulation that allows for the production of reports (products) communicating fire behavior and aviation hazard forecasts.

Research-to-Operations (R2O) Process

Objective: To transition the CAWFE® research capability into a robust operational decision support system (e.g., CO-FPS) that can meet the needs of a broad group of Colorado decision makers.

Testbed DefinitionA framework for conducting rigorous, transparent, and replicable testing of scientific theories and new technologies.

The Technology “Valley of Death”

The “Valley if Death” is where good research goes to die because of the complexity of the transition process and lack of funding opportunities.

CAWFE® has been an R&D effort for 20 years. The CO Legislature and the State recognized the potential value of this technology and is leading the effort to operationalize it.

Operational Use

Source: ieeecomputingsociety.org

Development Approach

To reduce risk and ensure that end-user requirements are met throughout the course of the project, the system will be developed using an iterative approach.

This approach is consistent with the overall concept that NCAR and the DFPC are in a partnership and will work cooperatively to develop and evolve the system capabilities as the user requirements and science and technology evolve.

Evolutionary Development Process

Design

IOC System

User

Requirements

Analysis

IOC

Coding &

Implementation

Develop a

Version

Incorporate

Customer

Feedback

Incorporate

Feedback

Elicit

Customer

Feedback

Elicit User

Feedback

Deliver

the

Version

Demonstrate

Annually

Deliver

Final

Version

Deliver Final

System

Initial Operating Capability DemoSep-Oct-Nov 2016

November2020

Initial Operating Capability CO-FPS Products

18 hour predictions (at user defined increments) of:

• Fire extent• Rate of spread• Heat release• Smoke concentration• Significant fire phenomena

• Turbulence intensity• Downdraft and updraft regions• Wind shear regions

• Wind speed, direction, gustiness• Surface air temperature• Surface relative humidity

Fire behavior product group

Aviation hazard product group

Fire weather product group

Planned System Attributes

• Real-time data ingest of weather, fuel, and active fire detection data from the MMA and Visible Infrared Imaging Radiometer Suite (VIIRS)

• Multiple fire model cycles (runs) per day (utilizing updated weather and fire mapping data)

• User ability to select fire prediction location and size (via CO-WIMS)

• User ability to input ignition information (via CO-WIMS)

• Output customized and formatted to be displayed on CO-WIMS

Rim fire in Central-East California. VIIRS active fire detection.

NASA, University of Maryland

MMA VIIRS

System Concept Diagram

Setting Expectations

• The IOC version of the CO-FPS will be limited and perhaps not 100% robust.

• User feedback gathered this year will be used to improve the system capabilities, features, functions, and performance.

• We look forward to working with you on this effort.

• The CO-FPS is not an off-the-shelf technology.

CAWFE® Description

Dr. Janice CoenMesoscale & Microscale Meteorology Laboratory

NCAR

Status of operational fire behavior models

• Weather is the wildcard in a wildland fire event.

• More sophisticated tools are emerging from research for practical use

• Current models (FARSITE, NearTerm fire) similarly: • Estimate how fast the leading edge of the fire will spread, based on separate

effects of wind speed, terrain, & fuel properties.• Use station measurements, simple approximations, or too coarse weather

forecasts• Do not include how the fire creates its own weather• Frequently require calibrating inputs to get observed fire behavior

• Current tools are weak in these (and other) events:

Plume-driven fires (fire-induced winds)

Mountain airflows Cloud gust fronts

Others:- Fire whirls

(Missionary Ridge)- Backfires (Spade

Fire)- Splitting/fire-

induced + wind-driven heads

- Chimney effect

NCAR Wildland Fire Modeling Science

The CAWFE® (Coupled Atmosphere-Wildland Fire Environment) modeling system couples a 4-D numerical weather prediction model designed for high resolution (100s of m) simulations in complex terrain with a wildland fire behavior model to predict fire weather, fire behavior, and fire-weather interactions.

CAWFE simulation of the 2014 King fire showing the heat release rate produced by the fire and near ground wind speed and direction.

CAWFE simulation:

The “universal” fire shape and fire whirls evolve

from fire-atmosphere interactions.

12

The Onion Fire, Owens Valley, CA (courtesy

Chuck George USFS)

To understand fire behavior fundamentals… … and the unfolding of wildfire events

Fire Behavior Module

Surface fire

2. Rate of spread

(ROS) of flaming

front calculated

as function of

fire-affected

wind, fuel, and

slope using

Rothermel

(1972) semi-

empirical

equations3. Post-frontal

heat & water

vapor release.

Once ignited, the

fuel remaining

decays

exponentially,

acc. to lab

experiments.

1. Represent &

track the (subgrid-

scale) interface

between burning

and nonburning

regions

(the‘flaming front’)

4. Heat, water

vapor, and smoke

released by

surface fire into

lowest layers of

atmospheric model

Overview of Components

Courtesy BLM

Fire Behavior Module

Surface fire

2. Rate of spread

(ROS) of flaming

front calculated

as function of

fire-affected

wind, fuel, and

slope using

Rothermel

(1972) semi-

empirical

equations3. Post-frontal

heat & water

vapor release.

Once ignited, the

fuel remaining

decays

exponentially, acc.

to lab

experiments.

1. Represent &

track the (subgrid-

scale) interface

between burning

and nonburning

regions

(the‘flaming front’)

4. Heat, water

vapor, and smoke

released by

surface fire into

lowest layers of

atmospheric model

Overview of Components

Crown fire

Courtesy BLM

5. Surface fire heats and

dries canopy. Does the

surface fire heat flux

exceed the (empirical)

threshold to transition

into the tree canopy (if

present)?

6. Calculate the rate of

spread of the crown

fire using empirical

relationships to

surface fire ROS

7. Heat, water vapor, and smoke released by

crown fire into atmospheric model

K. Cameron

CAWFE® Model Configuration

5 simultaneous nested weather modeling domains with horizontal grid spacing 10 km, 3.33 km, 1.11 km, 370 m, and 123 m telescope from a regional or national forecast…

INPUT DATA: (3) fuel model data (LANDFIRE)

…to, for example, a 25 km x 25 km area focusing on a fire.

Grid refinement

INPUT DATA: (1) Gridded synoptic/global weather analyses or NWS model forecast

INPUT DATA: (2) terrain elevation data

Wind speed in plane (in m s-1)

Simulated wind in vertical slice through ignition1 frame = 1 minute, beginning 5:45 a.m. 9 June

Height(m)

Distance W-E (m)

High Park Fire 6/9/12

X Ft.

Collins,

CO

N

Period: 6/9 5:30 am – 6/10 2:58 am (21.5 hr). Each frame: 1 min.

Animation: the rate the fire is releasing heat (inset color bar, W m-

2), burned area (dark brown), wind at 12 m (39 ft) above ground,

and smoke.

Distance W-E (m)

Yarnell Hill Fire Yarnell, AZ, 6/30/13

3:30 PM (C.Mass blog)

1 frame = 1 min

Testing and Verification Cases

• Big Elk Fire, CO • Hayman Fire, CO• Yarnell Hill Fire, AZ• King Fire, CA

CAWFE SIMULATION

Modeled weather, fire extent, shape, intensity, and land surface effects can be validated.Airborne or space borne infrared data reveal fire properties through smoke.

INFRARED DATAFireMapper, USDA Forest Service

ESPERANZA WILDFIRE

• Simi Fire, CA• Troy Fire, CA• Spade Fire, MT

Simulated large wildfires in many fuel & weather conditions:• Little Bear Fire, NM• High Park Fire, CO• Esperanza Fire, CA• Prototype real-time simulation of CO fires during 2004

Satellite Active Fire Detection

The new Suomi-NPP sensor VIIRS provides (at least) twice-daily fire detection data at resolutions relevant for fire behavior

(new) S-NPP/VIIRS 375m(old) Terra/MODIS 1km (old) Aqua/MODIS 1km

Wildfire in southern Brazil, March 2013

Courtesy W. Schroeder, Univ. of Maryland

VIIRS data can be used to start the fire ‘in progress’ and evaluate the prediction 12 h later

Yellow perimeter: VIIRS fire perimeter used for model initializationRed perimeter: VIIRS fire perimeter 12 h later

7.3 22.6 38.0

33.7

22.8

11.9x (km)

EXPTA

EXPTB

EXPTC

33.7

22.8

11.9

33.7

22.8

11.9

7.3 22.6 38.0

33.7

22.8

11.9

7.3 22.6 38.0

7.3 22.6 38.0

x (km)

x (km)

x (km)

Seq

uen

ce o

f m

od

el s

imu

lati

on

s

Time

Little Bear FirePurple: VIIRS Fire detection polygon

Red: simulated fire perimeter

Jun 8 2031 UTC Jun 9 0857 UTC Jun 9 2014 UTC Jun 10 0833 UTC

a) b) c)

e)

d)

f)

h)

g)

i)

j)

Coen and Schroeder, 2013, Geophysical Research Letters

Fire Prediction Use Cases

Initial Focus• High resolution fire weather products• Management of individual fires – fire behavior• Aviation weather hazard guidance (up/downdrafts,

rotors, wind shear, turbulence)

Additional Candidate Applications• Resource planning for regional operations• Support for prescribed fire planning and execution• Forest and rangeland management and planning• Smoke impacts and air quality• Firefighter safety & training• Forensic evaluations of fire ignition and evolution

Questions?

• Janice Coen (janicec@ucar.edu)

• Bill Mahoney (mahoney@ucar.edu)

• Jim Cowie (cowie@ucar.edu)